A power energy efficiency optimization method based on intelligent load management

Abstract

The present invention discloses a power energy efficiency optimization method based on intelligent load management, comprising the following steps: S1, generating an initial load feature matrix; S2, decomposing the initial load feature matrix into a short-term behavior feature matrix and a long-term trend feature matrix through pseudo-symmetry decoupling; S3, performing multi-level load demand prediction through a fusion structure of multi-scale convolution and a recurrent neural network; S4, encoding a cognitive gene sequence, triggering gene mutation and recombination operations when significant changes occur, and generating a load distribution strategy; S5, transmitting the load distribution strategy into a causal inference and self-supervised feedback control module, and triggering a self-learning mechanism if the deviation between the feedback data and the prediction result exceeds a threshold; S6, dynamically allocating load priorities to generate a decision path for load management; S7, performing adaptive optimization through a multi-layer feature feedback closed-loop optimization system. The present invention adopts an intelligent load management method to optimize power energy efficiency through multi-layer feature decoupling and closed-loop optimization.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent load management, and particularly to a power energy efficiency optimization method based on intelligent load management. Background Art

[0002] In the fields of intelligent power management and load regulation, the operation efficiency and energy efficiency optimization of power systems have long been key research topics. With the rapid growth of social electricity demand and the gradual popularization of renewable energy, the load of power systems shows great volatility and uncertainty, and traditional load management methods are difficult to meet the needs of modern power management. Generally, the main purpose of power system load management is to improve the utilization efficiency of power resources and system stability through load forecasting and priority adjustment on the basis of ensuring the balance between power supply and demand, so as to achieve the energy efficiency optimization of the system. However, the existing load management technologies still have significant deficiencies in aspects such as load forecasting accuracy, strategy self-adaptability, and dynamic response speed.

[0003] Most traditional power load management methods adopt regulation means based on experience or rules to analyze and predict elements such as electricity prices, load demands, and environmental changes. Such methods mainly rely on historical data and use simple mathematical models to predict short-term load demands and formulate load distribution strategies. However, due to the lack of flexible self-adaptability of traditional methods, they are slow to respond when dealing with rapid load fluctuations or sudden changes. At the same time, traditional load management systems mostly rely on fixed rules and thresholds and cannot adjust load strategies in real time according to the dynamic changes of actual situations, which limits the sensitivity of the system to load fluctuations and the processing ability for different load priorities. In this way, the energy efficiency of the power system is likely to decrease during high load peak periods, and even lead to system instability. In addition, traditional methods are difficult to effectively comprehensively consider the impacts of multi-dimensional factors such as electricity prices, load demand fluctuations, and response times on load management, resulting in inaccurate load forecasting, low utilization rate of power resources, and hindering the further improvement of power system energy efficiency.

[0004] In recent years, intelligent power management methods have made certain progress in improving load forecasting accuracy and strategy optimization. Some load forecasting methods based on machine learning and deep learning have gradually been applied to load management systems. Through in-depth mining of big data and historical load data, some methods can achieve high-precision load demand forecasting. However, these methods still face several key problems. First, load data has highly complex and non-linear characteristics, affected by various factors such as environmental conditions, electricity price fluctuations, and user behavior. Existing load forecasting methods have limited effects in feature decoupling and dynamic modeling, and it is difficult to accurately separate short-term fluctuations and long-term trends. Second, although some systems have introduced deep learning models, their real-time performance and self-adaptability in dealing with complex load scenarios are still insufficient. Model parameters cannot be updated at any time, resulting in the system being difficult to flexibly adjust load strategies in scenarios where demand changes rapidly.

[0005] On the other hand, in the generation and adjustment of load management strategies, traditional methods mostly rely on a single optimization algorithm and lack multi-dimensional decision support for strategy changes. For example, under conditions of large load fluctuations or frequent electricity price changes, existing technologies lack an effective strategy regulation mechanism and cannot dynamically optimize resource allocation according to the priority requirements of different load units. In addition, existing methods rarely introduce causal reasoning and feedback mechanisms, resulting in the system being unable to effectively identify external environmental disturbances during the implementation of load strategies and lacking means to eliminate external influences, which affects the accuracy and robustness of load distribution strategies. Traditional feedback mechanisms mainly rely on fixed thresholds or static feedback, and this method cannot adaptively adjust strategies, imposing significant limitations on the long-term load forecasting and distribution optimization of the system. When the system executes load management strategies, it is difficult to continuously and adaptively adjust and optimize load responses, thereby affecting the overall energy efficiency of the power system.

[0006] In terms of priority allocation in load management, existing technologies mostly adopt simple priority settings based on electricity prices or load demands, ignoring the interactive effects of response time, electricity price fluctuations, and demand fluctuations, resulting in difficulty in forming a dynamic allocation strategy that adapts to changing load demands in the actual load environment. The usual priority setting methods mainly use load demand or real-time electricity price as the main criteria and fail to make hierarchical decisions according to the multi-level requirements of load characteristics. As a result, high-priority loads may not be fully guaranteed, while low-priority loads may result in waste in resource allocation.

[0007] Therefore, how to provide a power energy efficiency optimization method based on intelligent load management is an urgent problem that needs to be solved by those skilled in the art. Summary of the Invention

[0008] An object of the present invention is to propose a power energy efficiency optimization method based on intelligent load management. The present invention adopts an intelligent load management method and realizes precise optimization of power energy efficiency through multi-layer feature decoupling and closed-loop optimization technology. First, the present invention decomposes the load characteristics into short-term behavior characteristics and long-term trend characteristics through a dynamic feature decoupling module, so as to ensure accurate prediction of load demands on different time scales; on this basis, a closed-loop feedback optimization mechanism is adopted to adjust the load distribution strategy in real time to improve the sensitivity and response speed to load fluctuations. The method of the present invention has the advantages of rapid response, high energy efficiency, and flexible allocation, and provides an effective solution for intelligent load management in a complex power environment.

[0009] A power energy efficiency optimization method based on intelligent load management according to an embodiment of the present invention includes the following steps:

[0010] S1. Collect load data and perform data preprocessing to generate an initial load feature matrix;

[0011] S2. Input the initial load feature matrix into a dynamic feature decoupling module driven by adaptive contrast learning, and decompose the initial load feature matrix into a short-term behavior feature matrix and a long-term trend feature matrix through pseudo-symmetry decoupling;

[0012] S3. Input the short-term behavior feature matrix and the long-term trend feature matrix into a multi-scale focused load prediction model, and perform multi-level load demand prediction through a fusion structure of multi-scale convolution and recurrent neural network, and output short-term load demand, medium-term load demand, and long-term load demand;

[0013] S4. Through a cognitive gene sequence generation module, encode load characteristics, load priorities, and historical policies into a cognitive gene sequence. When the environment or load demand changes significantly, trigger gene mutation and recombination operations to generate an adaptive load distribution strategy;

[0014] S5. Input the load distribution strategy into a causal reasoning and self-supervised feedback control module, identify the true characteristics of the load demand based on causal reasoning, eliminate external environmental interference factors, and compare the feedback data with the prediction results. If the deviation exceeds the threshold, trigger the self-learning mechanism;

[0015] S6. Through a hierarchical load decision network, dynamically allocate load priorities according to real-time electricity prices, load demand fluctuations, and response times to generate a decision path for current load management;

[0016] S7. Through a multi-layer feature feedback closed-loop optimization system, collect load execution data and environmental change data in real time, compare the actual execution effect with the prediction result, and perform adaptive optimization based on the feedback result.

[0017] Optionally, the S2 specifically includes:

[0018] S21. Input the preprocessed initial load feature matrix into the adaptive contrast learning-driven dynamic feature decoupling module, and the adaptive contrast learning-driven dynamic feature decoupling module normalizes the initial load feature matrix;

[0019] S22. Use the pseudo-symmetry decoupling method to decompose the initial load feature matrix and establish a joint representation model of the short-term behavior feature matrix and the long-term trend feature matrix ;

[0020] S23. Decompose the initial load feature matrix:

[0021] ;

[0022] wherein, , and represent coefficient parameters, , and represent exponential parameters, represents the time gain or attenuation coefficient, represents the upper limit of time integration, represents the frequency parameter of the short-term behavior feature, represents the frequency parameter of the long-term trend feature, represents a positive offset;

[0023] S24. Through the adaptive contrast learning mechanism, optimize the coupling effect of the short-term behavior feature matrix and the long-term trend feature matrix , and use the loss function of contrast learning to balance the similarity and difference between the short-term behavior feature matrix and the long-term trend feature matrix ;

[0024] S25. During the contrast learning process, refine the short-term behavior feature matrix and the long-term trend feature matrix . The short-term behavior feature matrix includes the high-frequency features and periodic fluctuations of short-term load changes, and the long-term trend feature matrix includes the long-term trend and low-frequency features;

[0025] S26. Output the finally optimized short-term behavior feature matrix and the long-term trend feature matrix .

[0026] Optionally, the specific content of S3 includes:

[0027] S31. Input the short-term behavior feature matrix into the multi-scale convolutional layer of the at-most multi-scale focused load prediction model, extract short-term features of different time scales through multi-scale convolutional kernels, use small-scale convolutional kernels to capture high-frequency components of rapid load fluctuations, and generate a preliminary short-term load demand prediction matrix;

[0028] S32. Perform multi-scale convolutional processing on the long-term trend feature matrix to extract low-frequency features of the long-term trend with large-scale convolutional kernels in the multi-layer structure, capture the overall trend of long-term load changes by expanding the receptive field of the convolutional kernel, and generate a preliminary long-term load demand prediction matrix;

[0029] S33. Transmit the generated preliminary short-term load demand prediction matrix to the recurrent neural network layer to capture the time dependence of short-term features, recursively analyze and gradually refine the time series characteristics of short-term load demand. Each layer of recursion is based on the short-term changes in the previous time step, and gradually optimize the preliminary short-term load demand prediction matrix to generate a final short-term load demand matrix that adapts to demand fluctuations within a short period of time;

[0030] S34. Perform recurrent neural network processing on the generated preliminary long-term load demand prediction matrix, based on the periodic trend of long-term load, extract the stable change pattern of load over a long time span through deep recursion, and generate a final long-term load demand matrix;

[0031] S35. Generate a preliminary medium-term load demand prediction matrix, combine the final short-term load demand prediction matrix and the final long-term load demand matrix into a medium-term input matrix. The preliminary medium-term load demand prediction matrix includes the dual characteristics of short-term fluctuations and long-term trends. Apply multi-scale convolutional operations to the fused preliminary medium-term load demand prediction matrix to extract medium-frequency load demand change features, and output a final medium-term load demand prediction matrix that adapts to periodic fluctuations and longer-term changes;

[0032] S36. Output the prediction results of the final short-term load demand matrix, the final long-term load demand matrix, and the final medium-term load demand prediction matrix, and perform comprehensive adjustment through a multi-level feature focusing mechanism. The multi-level feature focusing mechanism adaptively optimizes the prediction results according to the real-time volatility of load demand and historical trend data.

[0033] Optionally, the specific content of S4 includes:

[0034] S41. Encode the load characteristics, load priorities, and historical policy data into multi-level cognitive gene sequences. Each gene sequence includes feature markers, priority parameters, historical adjustment policies, and context dependencies. The gene sequences at each level are respectively used for short-term load demand, medium-term load demand, and long-term load demand;

[0035] S42. Construct an intelligent load management gene bank, and store the multi-level cognitive gene sequences under different load scenarios into the intelligent load management gene bank. The intelligent load management gene bank dynamically adjusts the sorting rules and storage priorities of gene units according to environmental conditions, the amplitude of load demand changes, and load priorities, and clusters the multi-level cognitive gene sequences to automatically identify frequently occurring load change patterns;

[0036] S43. When a significant change in load demand or external environmental conditions is detected, identify the change type and activate the condition-adaptive gene mutation operation. Based on load priorities, mutation probabilities, and context dependencies, adjust the policy weights, feature markers, and context dependencies in the gene unit to form new mutant gene units;

[0037] S44. Perform multi-level gene recombination operations. By fusing new mutant gene units with different-level gene units in the intelligent load management gene bank, form an adaptive load distribution strategy. During the gene recombination process, automatically select different-level policy features from the gene units at each level to generate an adaptive load distribution strategy;

[0038] S45. Apply the load distribution strategy to the current load management. Based on the comparison between real-time load response data and the target energy efficiency, if the strategy effect is lower than the preset threshold, re-trigger gene mutation and recombination, and retain the traces of the strategy effect in the intelligent load management gene bank;

[0039] S46. Add the highly efficient gene units that have been adaptively optimized to the core layer of the intelligent load management gene bank, and update the structure of the intelligent load management gene bank through a reinforcement learning mechanism. Weight the gene units according to the occurrence frequency and adaptive effects of different load scenarios, eliminate low-priority or unused gene units, and dynamically optimize the content of the intelligent load management gene bank.

[0040] Optionally, the specific steps of S5 include:

[0041] S51. Input the load distribution strategy into the causal inference module, construct a load demand causal graph, identify the main influencing features of the load demand, and eliminate the interference factors of the external environment. The causal inference module is based on the causal relationship between load characteristics, environmental parameters, and load responses, and extracts core features by analyzing causal effects;

[0042] S52. Calculate the causal contribution value of each influencing feature through causal effect analysis. The causal contribution value is used to quantify the actual impact of each feature on the load demand:

[0043] ;

[0044] Among them, represents the causal contribution value of feature , represents the weight coefficient of feature , represents the weight coefficient of feature , represents the output variable of the load demand, represents the modulation coefficient of the control feature on the demand impact, represents the modulation coefficient of the control feature on the demand impact, represents the frequency parameter of feature , represents the phase shift of feature , represents the frequency parameter of feature , represents the phase shift of feature , represents the upper limit of the time window, represents the total number of features, represents the small value offset;

[0045] S53. Input the load distribution strategy into the self-supervised feedback control module to monitor the deviation between the actual load execution effect and the prediction result in real time. The self-supervised feedback control module uses the self-supervised learning mechanism to adaptively adjust the load distribution strategy according to the actual load execution effect by analyzing the potential patterns in the load response data;

[0046] S54. Set a deviation threshold. When the deviation between the load execution result and the prediction result exceeds the threshold, trigger the self-learning mechanism, and the self-learning mechanism adjusts the parameters and causal structure of the load distribution strategy based on the deviation;

[0047] S55. Build an adaptive deviation correction model to dynamically adjust the control parameters according to the deviation magnitude and trend, and calculate the deviation correction signal:

[0048] ;

[0049] Among them, represents the deviation correction signal, represents the deviation signal between the load execution result and the prediction result, represents the exponent for deviation magnitude adjustment, and represents a frequency parameter, and represents a phase offset, represents an adjustment coefficient, represents the average value of the deviation signal, represents an adjustment index, represents an adjustment coefficient, represents a denominator offset, represents the number of modulation terms used for processing the deviation signal, represents the number of adjustment terms for processing the difference between the average value of the deviation signal and the current deviation signal;

[0050] S56. Re - input the feedback - adjusted load distribution strategy into the causal inference module, evaluate the contribution degree of each feature based on the updated load demand causal graph, eliminate or optimize the strategy parameters that do not adapt to the current environment, and continuously update the parameters of the causal inference and self - supervised feedback control modules through a double - loop mechanism to achieve dynamic adaptive optimization of the load management strategy.

[0051] Optionally, the specific steps of S6 are as follows:

[0052] S61. Construct a hierarchical load decision - making network, input the real - time electricity price, load demand fluctuation, and response time as input features into the dynamic allocation layer of the network, and initially divide the priority factors of load units. The hierarchical load decision - making network adjusts the response strategy of each load unit in real - time according to the load demand change and electricity price fluctuation;

[0053] S62. In the dynamic allocation layer, analyze the current values of the real - time electricity price, load demand fluctuation, and response time to generate load priority factors:

[0054] ;

[0055] wherein, represents the priority factor of load unit ; represents the weight for adjusting the priority during high - demand fluctuations, represents the change rate of electricity price response, represents the high - priority threshold of the electricity price, represents the demand fluctuation weight under medium - demand fluctuation conditions, represents the electricity price smoothing coefficient, represents the median value of the electricity price, represents the response change rate of medium - electricity - price fluctuations, represents the weight for adjusting low - priority loads, represents the weakening factor for controlling the response time of low - priority loads, Represents the low - priority value of the electricity price, Represents the response intensity of the low - priority load electricity price fluctuation, Represents the high - priority threshold of the response time, Represents the maximum - priority threshold of the response time, Represents the high - fluctuation threshold of the load demand fluctuation, Represents the low - fluctuation threshold of the load demand fluctuation, Represents the load unit 's real - time electricity price, Represents the load unit 's load demand fluctuation, Represents the load unit 's response time;

[0056] S63. Classify the load units according to the priority factor, and divide the load into high - priority, medium - priority and low - priority; the high - priority load units are the loads that are sensitive to electricity price fluctuations or give priority to response in the case of large demand fluctuations; the medium - priority load units are used for relatively important and flexibly adjustable loads; the low - priority load units are non - critical loads with long response time and small load fluctuations.

[0057] S64. Generate a dynamic load management path in the hierarchical load decision - making network, give priority to responding to high - priority load units, process medium - priority load units when the electricity price is medium or the demand is stable, and arrange low - priority load units to be executed during low - electricity - price periods.

[0058] S65. Monitor the execution of the load management path. If the changes in the real - time electricity price, load demand fluctuation or response time detected during the path execution exceed the preset threshold, recalculate the priority factor of each load unit and dynamically adjust the load management path.

[0059] The beneficial effects of the present invention are:

[0060] First of all, through the dynamic feature decoupling module driven by adaptive contrast learning, the present invention decomposes the load data into short - term behavior features and long - term trend features, and can accurately extract the short - term fluctuations and long - term changes of the load demand. This feature decoupling method enables the load management system to effectively separate the different time characteristics of the load in a complex load fluctuation environment, thereby providing a more accurate data basis for multi - scale load forecasting. Combined with the multi - scale focused load forecasting model, the present invention can respectively predict the high - frequency changes of short - term load demand, medium - frequency fluctuations of medium - term demand and long - term load trends in a hierarchical manner, effectively avoiding the limitation that traditional models are difficult to accurately capture load fluctuations on a single time scale.

[0061] Through the cognitive gene sequence generation module and the causal reasoning and self-supervised feedback control module, the load management system of the present invention has high efficiency and adaptability. When the load demand changes or the environmental conditions fluctuate significantly, the system can quickly generate a load distribution strategy suitable for the current situation through the cognitive gene mutation and recombination mechanism, significantly improving the system's response speed. The causal reasoning module can identify the main influencing features in the load demand, eliminate environmental interference factors, and ensure the accuracy and robustness of the load strategy. At the same time, the self-supervised feedback control module realizes the closed-loop optimization of the system by monitoring the deviation between the execution effect and the prediction result in real time, enabling the load management strategy to continuously self-adjust and improve during actual execution, thereby ensuring the efficiency and dynamic adaptability of load distribution.

[0062] In addition, the hierarchical load decision-making network proposed by the present invention dynamically allocates load priorities through factors such as real-time electricity prices, load demand fluctuations, and response times, thus effectively solving the problem of single priority setting in the prior art. The system dynamically adjusts the priority factor according to different load characteristics through a piecewise function formula, forming a hierarchical response mechanism with high, medium, and low three levels of priorities. High-priority load units respond first in case of emergencies, while medium- and low-priority loads are arranged for execution according to real-time electricity prices and demand fluctuations respectively, reducing the system operation cost while ensuring the reasonable allocation of load resources. Through this multi-dimensional decision-making network, the present invention realizes flexible response to load demand. Even under complex load fluctuation conditions, the system can still operate stably and achieve the optimization effect of energy efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0063] The drawings are used to provide a further understanding of the present invention, and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the present invention, but do not constitute a limitation to the present invention. In the drawings:

[0064] Figure 1 is a flowchart of a power energy efficiency optimization method based on intelligent load management proposed by the present invention;

[0065] Figure 2 is a schematic structural diagram of a multi-scale focused load prediction model of a power energy efficiency optimization method based on intelligent load management proposed by the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0066] Now, the present invention will be further described in detail with reference to the accompanying drawings. These drawings are all simplified schematic diagrams, only illustrating the basic structure of the present invention in a schematic manner, so they only show the components related to the present invention.

[0067] Refer to Figure 1 and Figure 2 , a power energy efficiency optimization method based on intelligent load management, includes the following steps:

[0068] S1. Collect load data and perform data preprocessing to generate an initial load feature matrix;

[0069] S2. Input the initial load feature matrix into the dynamic feature decoupling module driven by adaptive contrast learning, and decompose the initial load feature matrix into a short-term behavior feature matrix and a long-term trend feature matrix through pseudo-symmetry decoupling;

[0070] S3. Input the short-term behavior feature matrix and the long-term trend feature matrix into the multi-scale focused load prediction model, and perform multi-level load demand prediction through the fusion structure of multi-scale convolution and recurrent neural network, and output short-term load demand, medium-term load demand, and long-term load demand;

[0071] S4. Through the cognitive gene sequence generation module, encode the load characteristics, load priorities, and historical policies into a cognitive gene sequence. When the environment or load demand changes significantly, trigger gene mutation and recombination operations to generate an adaptive load allocation strategy;

[0072] S5. Input the load allocation strategy into the causal reasoning and self-supervised feedback control module, identify the true characteristics of the load demand based on causal reasoning, eliminate external environmental interference factors, and compare the feedback data with the prediction results. If the deviation exceeds the threshold, trigger the self-learning mechanism;

[0073] S6. Through the hierarchical load decision-making network, dynamically allocate load priorities according to the real-time electricity price, load demand fluctuations, and response time, and generate the decision-making path for current load management;

[0074] S7. Through the multi-layer feature feedback closed-loop optimization system, collect load execution data and environmental change data in real time, compare the actual execution effect with the prediction results, and perform adaptive optimization based on the feedback results.

[0075] In this embodiment, the S2 specifically includes:

[0076] S21. Input the preprocessed initial load feature matrix into the dynamic feature decoupling module driven by adaptive contrast learning, and the dynamic feature decoupling module driven by adaptive contrast learning performs normalization processing on the initial load feature matrix;

[0077] S22. Use the pseudo-symmetry decoupling method to perform feature decomposition on the initial load feature matrix, and establish a joint representation model of the short-term behavior feature matrix and the long-term trend feature matrix ;

[0078] S23. Decompose the initial load feature matrix:

[0079] ;

[0080] Among them, 、 and represent coefficient parameters, 、 and represent exponential parameters, represents the time gain or attenuation coefficient, represents the upper limit of time integration, represents the frequency parameter of short-term behavior characteristics, represents the frequency parameter of long-term trend characteristics, represents a positive offset;

[0081] S24. Through the adaptive contrast learning mechanism, optimize the coupling effect of the short-term behavior feature matrix and the long-term trend feature matrix . Use the loss function of contrast learning to balance the similarity and difference between the short-term behavior feature matrix and the long-term trend feature matrix ;

[0082] S25. During the contrast learning process, refine the short-term behavior feature matrix and the long-term trend feature matrix . The short-term behavior feature matrix includes high-frequency features and periodic fluctuations of short-term load changes, and the long-term trend feature matrix includes long-term trends and low-frequency features;

[0083] S26. Output the finally optimized short-term behavior feature matrix and the long-term trend feature matrix .

[0084] In this embodiment, the specific content of S3 is as follows:

[0085] S31. Input the short-term behavior feature matrix into the multi-scale convolution layer of the multi-scale focused load prediction model, extract short-term features of different time scales through multi-scale convolution kernels, and use small-scale convolution kernels to capture the high-frequency components of rapid load fluctuations to generate a preliminary short-term load demand prediction matrix;

[0086] S32. Perform multi-scale convolution processing on the long-term trend feature matrix . Extract the low-frequency features of the long-term trend through the large-scale convolution kernels in the multi-layer structure, and capture the overall trend of long-term load changes by expanding the receptive field of the convolution kernels to generate a preliminary long-term load demand prediction matrix;

[0087] S33. Transmit the generated preliminary short-term load demand prediction matrix to the recurrent neural network layer to capture the time dependence of short-term features, and recursively analyze to gradually refine the time series characteristics of short-term load demand. Each layer of recursion is based on the short-term changes in the previous time step, gradually optimizing the preliminary short-term load demand prediction matrix to generate the final short-term load demand matrix that adapts to the demand fluctuations within a short period of time;

[0088] S34. Perform recurrent neural network processing on the generated preliminary long-term load demand prediction matrix. Based on the periodic trend of long-term load, extract the stable change pattern of load over a long time span through deep recursion to generate the final long-term load demand matrix;

[0089] S35. Generate the preliminary medium-term load demand prediction matrix, combine the final short-term load demand prediction matrix and the final long-term load demand matrix into a medium-term input matrix. The preliminary medium-term load demand prediction matrix includes the dual characteristics of short-term fluctuations and long-term trends. Apply multi-scale convolution operations to the fused preliminary medium-term load demand prediction matrix to extract the load demand change characteristics of medium frequency, and output the final medium-term load demand prediction matrix that adapts to periodic fluctuations and longer-term changes;

[0090] S36. Output the prediction results of the final short-term load demand matrix, the final long-term load demand matrix, and the final medium-term load demand prediction matrix, and perform comprehensive adjustment through a multi-level feature focusing mechanism. The multi-level feature focusing mechanism adaptively optimizes the prediction results according to the real-time volatility of load demand and historical trend data.

[0091] In this embodiment, the specific content of S4 includes:

[0092] S41. Encode load characteristics, load priorities, and historical policy data into multi-level cognitive gene sequences. Each gene sequence includes feature markers, priority parameters, historical adjustment policies, and context dependencies. The gene sequences at each level are respectively used for short-term load demand, medium-term load demand, and long-term load demand;

[0093] S42. Construct an intelligent load management gene bank, store the multi-level cognitive gene sequences in different load scenarios into the intelligent load management gene bank. The intelligent load management gene bank dynamically adjusts the sorting rules and storage priorities of gene units according to environmental conditions, the amplitude of load demand changes, and load priorities, and clusters the multi-level cognitive gene sequences to automatically identify the frequently occurring load change patterns;

[0094] S43. When a significant change in load demand or external environmental conditions is detected, identify the type of change and activate the condition - adaptive gene mutation operation. Based on load priority, mutation probability, and context - dependence, adjust the policy weights, feature markers, and context - dependence in the gene unit to form a new mutant gene unit.

[0095] S44. Perform a multi - level gene recombination operation. By fusing the new mutant gene unit with different - level gene units in the intelligent load management gene bank, form an adaptive load distribution strategy. During the gene recombination process, automatically select different - level strategy features from gene units at each level to generate an adaptive load distribution strategy.

[0096] S45. Apply the load distribution strategy to the current load management. Based on the comparison between real - time load response data and the target energy efficiency, if the strategy effect is lower than the preset threshold, re - trigger gene mutation and recombination, and retain the trace of the strategy effect in the intelligent load management gene bank.

[0097] S46. Add the highly efficient gene unit after adaptive optimization to the core layer of the intelligent load management gene bank, and update the structure of the intelligent load management gene bank through a reinforcement learning mechanism. Weight the gene units according to the occurrence frequency and adaptive effect of different load scenarios, eliminate low - priority or unused gene units, and dynamically optimize the content of the intelligent load management gene bank.

[0098] In this embodiment, the specific steps of S5 are as follows:

[0099] S51. Input the load distribution strategy into the causal inference module, construct a load - demand causal graph, identify the main influencing features of the load demand, and eliminate the interference factors of the external environment. The causal inference module is based on the causal relationship between load characteristics, environmental parameters, and load response, and extracts core features by analyzing causal effects.

[0100] S52. Calculate the causal contribution value of each influencing feature through causal effect analysis. The causal contribution value is used to quantify the actual impact of each feature on the load demand:

[0101] ;

[0102] Among them, represents the causal contribution value of feature , represents the weight coefficient of feature , represents the weight coefficient of feature , represents the output variable of the load demand, represents the modulation coefficient of the control feature on the demand impact. Represents a control feature Modulation coefficient for demand impact, Represents a feature Frequency parameter of, Represents a feature Phase offset of, Represents a feature Frequency parameter of, Represents a feature Phase offset of, Represents the upper limit of the time window, Represents the total number of features, Represents the small value offset;

[0103] S53. Pass the load distribution strategy into the self-supervised feedback control module to monitor the deviation between the actual load execution effect and the prediction result in real time. The self-supervised feedback control module uses a self-supervised learning mechanism to adaptively adjust the load distribution strategy according to the actual load execution effect by analyzing the potential patterns in the load response data;

[0104] S54. Set a deviation threshold. When the deviation between the load execution result and the prediction result exceeds the threshold, trigger the self-learning mechanism, and the self-learning mechanism adjusts the parameters and causal structure of the load distribution strategy based on the deviation;

[0105] S55. Build an adaptive deviation correction model to dynamically adjust the control parameters according to the deviation magnitude and trend, and calculate the deviation correction signal:

[0106] ;

[0107] Wherein, Represents the deviation correction signal, Represents the deviation signal between the load execution result and the prediction result, Represents the exponent for deviation magnitude adjustment, And Represents the frequency parameter, And Represents the phase offset, Represents the adjustment coefficient, Represents the average value of the deviation signal, Represents the adjustment exponent, Represents the adjustment coefficient, Represents the denominator offset, Represents the number of modulation terms used for processing the deviation signal, Represents the number of adjustment terms for processing the difference between the average value of the deviation signal and the current deviation signal;

[0108] S56. Re - input the load distribution strategy adjusted by feedback into the causal reasoning module, evaluate the contribution degrees of each feature based on the updated causal graph of load demand, eliminate or optimize the policy parameters that do not adapt to the current environment, and continuously update the parameters of the causal reasoning and self - supervised feedback control modules through a double - loop mechanism to achieve the dynamic adaptive optimization of the load management strategy.

[0109] In this embodiment, the specific steps of S6 are as follows:

[0110] S61. Construct a hierarchical load decision - making network, input real - time electricity price, load demand fluctuation, and response time as input features into the dynamic allocation layer of the network, initially divide the priority factors of load units, and the hierarchical load decision - making network adjusts the response strategy of each load unit in real - time according to the changes in load demand and electricity price fluctuations;

[0111] S62. In the dynamic allocation layer, analyze the current values of real - time electricity price, load demand fluctuation, and response time to generate load priority factors:

[0112] ;

[0113] Among them, represents the priority factor of load unit , represents the weight for adjusting the priority during high - demand fluctuations, represents the change rate of electricity - price response, represents the high - priority threshold of electricity price, represents the demand - fluctuation weight under medium - demand - fluctuation conditions, represents the electricity - price smoothing coefficient, represents the median value of electricity price, represents the response change rate for medium - electricity - price fluctuations, represents the weight for adjusting low - priority loads, represents the weakening factor for controlling the response time of low - priority loads, represents the low - priority value of electricity price, represents the response intensity of low - priority - load electricity - price fluctuations, represents the high - priority threshold of response time, represents the maximum - priority threshold of response time, represents the high - fluctuation threshold of load demand fluctuation, represents the low - fluctuation threshold of load demand fluctuation, represents the load unit 's real - time electricity price, represents the load unit 's load demand fluctuation, represents the load unit 's response time;

[0114] S63. Classify the load units according to the priority factor, and divide the load into high priority, medium priority and low priority; the high-priority load units are the loads that are sensitive to electricity price fluctuations or give priority to response under large demand fluctuations; the medium-priority load units are used for relatively important and flexibly adjustable loads; the low-priority load units are non-critical loads with long response time and small load fluctuations.

[0115] S64. Generate a dynamic load management path in the hierarchical load decision network, give priority to responding to high-priority load units, process medium-priority load units when the electricity price is medium or the demand is stable, and arrange low-priority load units to be executed during low electricity price periods.

[0116] S65. Monitor the execution of the load management path. If it is detected during the path execution that the changes in real-time electricity price, load demand fluctuations or response time exceed the preset threshold, recalculate the priority factors of each load unit and dynamically adjust the load management path.

[0117] Embodiment 1:

[0118] To verify the feasibility of the present invention in implementation, the present invention is applied to the power management of a large commercial park. Due to the huge fluctuations in electricity demand over time and weather, the instability of the power load has become the main problem. Especially during peak electricity consumption periods and in extreme weather conditions, the power load fluctuates greatly, posing challenges to the stability and energy efficiency management of the power system. Traditional load management methods cannot flexibly cope with complex load fluctuations and usually only use historical data and static rules for simple prediction and allocation, resulting in unreasonable allocation of power resources and low energy efficiency under peak loads. To solve this problem, this embodiment introduces a power energy efficiency optimization method based on intelligent load management, and through the application of multi-layer feature decoupling, multi-scale focused load prediction model, cognitive gene sequence generation, and hierarchical load decision network, realizes the accurate prediction and intelligent allocation optimization of the load demand in this large commercial park.

[0119] In this embodiment, we decompose the power load demand in the park into short-term behavior characteristics and long-term trend characteristics, and input these characteristics into a multi-scale focused load forecasting model. The load characteristic decoupling module generates a short-term behavior characteristic matrix and a long-term trend characteristic matrix respectively according to the short-term behavior and long-term changes of load fluctuations, ensuring that the load management system can accurately predict the load demand under the conditions of intense short-term load fluctuations and slow long-term trend changes. After the data preprocessing is completed, the short-term and long-term characteristic matrices are respectively input into the multi-scale focused load forecasting model. The multi-scale convolutional layer is used to capture high-frequency and low-frequency load characteristics, and the recurrent neural network layer is used to gradually extract time-dependent characteristics to output short-term, medium-term, and long-term load demand forecasts.

[0120] On this basis, the intelligent load management system adopts a cognitive gene sequence generation module, which encodes the priority, feature markers, and historical strategies of load units through cognitive genes to adapt to the real-time changes of load demand in the park. During the actual operation process, when the load demand or environmental conditions change significantly, the system triggers the cognitive gene mutation and recombination mechanism, and quickly generates an adaptive load distribution strategy according to the current environment to ensure the stable distribution of the system load when the demand suddenly changes.

[0121] To verify the effectiveness of the system, we conducted a field test of load management during a 4-month winter and spring transition period, and collected and compared the load distribution and energy efficiency performance of this system and the traditional load management system in the park. The experimental results show that the response speed of this intelligent load management system during peak hours has been significantly improved, and the energy efficiency has increased by about 13.8%. For example, during the typical peak load period in December, the total load demand in this park once reached 1,650 kilowatts. Under the accurate prediction and optimized distribution of the intelligent load management system, the distribution deviation of the peak load remained within 3%, while the distribution deviation of the traditional system reached 12%, which is significantly lower than that of the system of the present invention.

[0122] Through the hierarchical load decision-making network, we further perform dynamic priority allocation for the load. Especially in the case of real-time electricity price fluctuations and surging load demands, the system responds hierarchically according to the priorities of different load units. On the premise of ensuring the stable supply of high-priority loads, by adjusting the execution time periods of low-priority loads, non-critical loads are executed during the low electricity price period, thus achieving the effect of saving electricity costs. For example, on a certain day in mid-January, due to the low temperature weather, the load demand of the heating equipment in the park soared to 1,200 kilowatts, of which the high-priority heating load reached 700 kilowatts. The system preferentially allocated sufficient electricity according to the high-priority load demand, and postponed some non-critical loads such as landscape lighting and non-emergency water supply to the low electricity price period, realizing the smooth optimization of energy efficiency. Compared with the traditional system, the system of the present invention is more balanced in the distribution of peak loads and valley loads, increasing the overall power utilization efficiency of the park by about 15%.

[0123] In the feedback and adjustment stage, the system, through the self-supervised feedback control module, monitors in real time the deviation between the load prediction result and the actual load demand. When it detects that the deviation exceeds the set threshold, it automatically triggers the self-learning mechanism to update the parameters of the prediction model. For example, on a certain day in March, the predicted mid-term load demand in the park was 850 kilowatts, while the actual load demand reached 910 kilowatts, with a deviation of about 7%. After the system identified that the load prediction deviation exceeded the threshold, it automatically optimized and adjusted the weight parameters of the short-term and mid-term load prediction models, improving the subsequent load prediction accuracy by about 8% and significantly reducing the waste of power resources.

[0124] The intelligent load management method of the present invention has shown significant superiority in the actual application in the park. Through the coordinated action of modules such as multi-layer feature decoupling, multi-scale load prediction, and dynamic priority allocation, the system responds quickly during peak loads, ensuring the stable supply of high-priority loads. At the same time, during valley loads, it reasonably distributes non-critical loads, optimizing the power usage efficiency of the park. Especially for periods of seasonal alternation and large changes in load demands, the energy efficiency improvement of the system is particularly obvious, providing an effective solution for the power management of the park.

[0125] Table 1 Comparison table of operation data between traditional system and intelligent load management system

[0126] ;

[0127] Table 2 Data table of load demand and allocation priority

[0128] ;

[0129] In this experiment, the data in Table 1 above revealed significant differences between the intelligent load management system and the traditional system in terms of load distribution and energy efficiency performance. The traditional system had a large deviation in peak load management, with the deviation range between 10% and 14%, resulting in unstable energy efficiency performance. In contrast, the intelligent load management system had a significantly reduced load deviation during the same peak period, remaining between 3% and 4%, with an average deviation of only 3.325%. From the perspective of energy efficiency indicators, the average energy efficiency of the traditional system was 84.25 kWh / kilowatt-hour, while that of the intelligent system reached 95.25 kWh / kilowatt-hour, with an overall energy efficiency improvement of 13.2%. This difference indicates that the intelligent load management system performs excellently in real-time load regulation and resource optimization, effectively reducing the deviation rate of peak load management and improving the system energy efficiency.

[0130] Further analyzing the load demand and allocation priority data in Table 2, it can be seen that the intelligent load management system has obvious advantages in ensuring high-priority load demands. Taking the middle of December as an example, the high-priority load demand detected by the system was 700 kW, and this priority load was fully met in the actual allocation; the medium- and low-priority loads were also relatively close to the demand, with only minor deviations. This refined load allocation optimization effect was repeatedly verified in the experimental data for January, February, and March. The intelligent load management system arranges non-critical loads to be executed during periods with lower electricity prices based on real-time electricity prices and load fluctuations, minimizing the load pressure during peak periods. Overall, the dynamic load priority allocation strategy of the intelligent system significantly improves the utilization efficiency of electric power resources, can flexibly allocate loads of each priority level, and realizes the stable operation of the system under different load demand conditions.

[0131] Thus, it can be seen that the data in Table 1 and Table 2 further verify the superior performance of the intelligent load management system in a complex power environment. This system can effectively reduce the peak load deviation, improve the system energy efficiency, and ensure the demands of high-priority loads in load distribution, and meet the execution arrangements of medium- and low-priority loads in different electricity price periods. This hierarchical load priority decision-making method enables the system to exhibit excellent flexibility and stability in practical applications, providing strong technical support for efficient power energy efficiency optimization.

[0132] As described above, it is only the preferred specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art within the technical scope disclosed by the present invention, according to the technical solution and inventive concept of the present invention, makes equivalent substitutions or changes, and all should be covered within the protection scope of the present invention.

Claims

1. A power energy efficiency optimization method based on intelligent load management, characterized in that, It includes the following steps: S1. Collect load data and perform data preprocessing to generate an initial load feature matrix; S2. Input the initial load feature matrix into the dynamic feature decoupling module driven by adaptive contrast learning, and decompose the initial load feature matrix into a short-term behavior feature matrix and a long-term trend feature matrix through pseudo-symmetry decoupling; S3. Input the short-term behavior feature matrix and the long-term trend feature matrix into the multi-scale focused load forecasting model, and perform multi-level load demand forecasting through the fusion structure of multi-scale convolution and recurrent neural network, and output short-term load demand, medium-term load demand and long-term load demand; S4. Through the cognitive gene sequence generation module, encode the load characteristics, load priorities and historical policies into cognitive gene sequences. When the environment or load demand changes significantly, trigger gene mutation and recombination operations to generate adaptive load distribution strategies; S5. Input the load distribution strategy into the causal reasoning and self-supervised feedback control module, identify the true characteristics of the load demand based on causal reasoning, eliminate external environmental interference factors, and compare the feedback data with the prediction results. If the deviation exceeds the threshold, trigger the self-learning mechanism; S6. Through the hierarchical load decision-making network, dynamically allocate load priorities according to the real-time electricity price, load demand fluctuations and response time, and generate the decision-making path for current load management; S7. Through the multi-layer feature feedback closed-loop optimization system, collect load execution data and environmental change data in real time, compare the actual execution effect with the prediction result, and perform adaptive optimization based on the feedback result; The specific content of S2 includes: S21. Input the preprocessed initial load feature matrix into the dynamic feature decoupling module driven by adaptive contrast learning, and the dynamic feature decoupling module driven by adaptive contrast learning performs normalization processing on the initial load feature matrix; S22. Use the pseudo-symmetry decoupling method to perform eigen-decomposition on the initial load feature matrix, and establish a joint representation model for the short-term behavior feature matrix and the long-term trend feature matrix ; S23. Decompose the initial load feature matrix: ; Among them, , and represent coefficient parameters, , and represent exponent parameters, represents the time gain or attenuation coefficient, represents the upper limit of time integration, represents the frequency parameter of short-term behavior characteristics, represents the frequency parameter of long-term trend characteristics, represents a positive offset; S24. Through the adaptive contrastive learning mechanism, optimize the coupling effect of the short-term behavior feature matrix and the long-term trend feature matrix . Use the loss function of contrastive learning to balance the similarity and difference between the short-term behavior feature matrix and the long-term trend feature matrix ; S25. During the contrastive learning process, refine the short-term behavior feature matrix and the long-term trend feature matrix . The short-term behavior feature matrix includes high-frequency features and periodic fluctuations of short-term load changes, and the long-term trend feature matrix includes long-term trends and low-frequency features; S26. Output the finally optimized short-term behavior feature matrix and long-term trend feature matrix ; The specific content of S4 includes: S41. Encode the load characteristics, load priorities and historical policy data into multi-level cognitive gene sequences. Each gene sequence includes feature markers, priority parameters, historical adjustment strategies and context dependencies. The gene sequences at each level are respectively used for short-term load demand, medium-term load demand and long-term load demand; S42. Build an intelligent load management gene bank, store the multi-level cognitive gene sequences in different load scenarios into the intelligent load management gene bank. The intelligent load management gene bank dynamically adjusts the sorting rules and storage priorities of gene units according to environmental conditions, the change range of load demand and load priorities, and clusters the multi-level cognitive gene sequences to automatically identify frequently occurring load change patterns; S43. When it is detected that the load demand or external environmental conditions change significantly, identify the change type and activate the conditional adaptive gene mutation operation. Based on the load priority, mutation probability and context dependency, adjust the policy weights, feature markers and context dependencies in the gene unit to form new mutant gene units; S44. Perform multi-level gene recombination operations. By fusing new mutant gene units with different-level gene units in the intelligent load management gene pool, an adaptive load distribution strategy is formed. During the gene recombination process, different-level strategy features are automatically selected from the gene units at each level to generate an adaptive load distribution strategy; S45. Apply the load distribution strategy to the current load management. Based on the comparison between real-time load response data and the target energy efficiency, if the strategy effect is lower than the preset threshold, gene mutation and recombination are re-triggered, and the traces of the strategy effect are retained in the intelligent load management gene pool; S46. Add the highly efficient gene units that have been adaptively optimized to the core layer of the intelligent load management gene pool, and update the structure of the intelligent load management gene pool through a reinforcement learning mechanism. Weigh the gene units according to the occurrence frequency and adaptive effect of different load scenarios, eliminate low-priority or unused gene units, and dynamically optimize the content of the intelligent load management gene pool.

2. The power energy efficiency optimization method based on intelligent load management according to claim 1, wherein The specific steps of S3 are as follows: S31. Input the short-term behavior feature matrix into the multi-scale convolutional layer of the load prediction model with multi-scale focus, extract short-term features of different time scales through multi-scale convolutional kernels, use small-scale convolutional kernels to capture the high-frequency components of rapid load fluctuations, and generate a preliminary short-term load demand prediction matrix; S32. Perform multi-scale convolution processing on the long-term trend feature matrix to extract the low-frequency features of the long-term trend using the large-scale convolution kernels in the multi-layer structure. By expanding the receptive field of the convolution kernels, capture the overall trend of the long-term change in the load and generate a preliminary long-term load demand prediction matrix; S33. Transmit the generated preliminary short-term load demand prediction matrix to the recurrent neural network layer to capture the time dependence of short-term features. Through recursive analysis, gradually refine the time series characteristics of the short-term load demand. Each layer of recursion is based on the short-term changes in the previous time step, and gradually optimize the preliminary short-term load demand prediction matrix to generate the final short-term load demand prediction matrix that adapts to the demand fluctuations within a short period; S34. Perform recurrent neural network processing on the generated preliminary long-term load demand prediction matrix. Based on the periodic trend of the long-term load, extract the stable change pattern of the load over a long time span through deep recursion to generate the final long-term load demand matrix; S35. Generate a preliminary medium-term load demand prediction matrix. Combine the final short-term load demand prediction matrix and the final long-term load demand matrix into a medium-term input matrix. The preliminary medium-term load demand prediction matrix includes the dual characteristics of short-term fluctuations and long-term trends. Apply multi-scale convolution operations to the fused preliminary medium-term load demand prediction matrix to extract the load demand change characteristics at medium frequencies, and output the final medium-term load demand prediction matrix that adapts to periodic fluctuations and longer-term changes; S36. Output the prediction results of the final short-term load demand matrix, the final long-term load demand matrix, and the final medium-term load demand prediction matrix, and perform comprehensive adjustment through a multi-level feature focusing mechanism. The multi-level feature focusing mechanism adaptively optimizes the prediction results according to the real-time volatility of the load demand and the historical trend data.

3. A power energy efficiency optimization method based on intelligent load management according to claim 1, characterized in that, The specific steps of S5 are as follows: S51. Input the load distribution strategy into the causal inference module to construct a load demand causal graph, identify the main influencing features of the load demand, and eliminate the interference factors of the external environment. The causal inference module is based on the causal relationship between load characteristics, environmental parameters, and load response, and extracts core features through causal effect analysis; S52. Calculate the causal contribution value of each influencing feature through causal effect analysis. The causal contribution value is used to quantify the actual impact of each feature on the load demand: ； Among them, represents the causal contribution value of feature ; represents the weight coefficient of feature ; represents the weight coefficient of feature ; represents the output variable of load demand, represents the modulation coefficient of the influence of control feature on the demand, represents the modulation coefficient of the influence of control feature on the demand, represents the frequency parameter of feature ; represents the phase shift of feature ; represents the frequency parameter of feature ; represents the phase shift of feature ; represents the upper limit of the time window, represents the total number of features, represents the small value offset; S53. Pass the load distribution strategy into the self-supervised feedback control module to monitor the deviation between the load execution effect and the prediction result in real time. The self-supervised feedback control module uses a self-supervised learning mechanism to adaptively adjust the load distribution strategy according to the actual load execution effect by analyzing the potential patterns in the load response data; S54. Set a deviation threshold. When the deviation between the load execution result and the prediction result exceeds the threshold, trigger the self-learning mechanism, which adjusts the parameters and causal structure of the load distribution strategy based on the deviation; S55. Construct an adaptive deviation correction model to dynamically adjust the control parameters according to the deviation magnitude and change trend, and calculate the deviation correction signal: ； Among them, represents the deviation correction signal, represents the deviation signal between the load execution result and the predicted result, represents the exponent for deviation magnitude adjustment, and represents the frequency parameter, and represents the phase shift, represents the adjustment coefficient, represents the average value of the deviation signal, represents the adjustment exponent, represents the adjustment coefficient, represents the denominator offset, represents the number of modulation terms used for processing the deviation signal, represents the number of adjustment terms for processing the difference between the average value of the deviation signal and the current deviation signal; S56. Re-pass the load distribution strategy adjusted by feedback into the causal reasoning module, evaluate the contribution degree of each feature based on the updated load demand causal graph, eliminate or optimize the policy parameters that do not adapt to the current environment, and continuously update the parameters of the causal reasoning and self-supervised feedback control modules through a double-loop mechanism to achieve dynamic adaptive optimization of the load management strategy.

4. A power energy efficiency optimization method based on intelligent load management according to claim 1, characterized in that, The specific steps of S6 are as follows: S61. Construct a hierarchical load decision-making network. Take the real-time electricity price, load demand fluctuation, and response time as input features and pass them into the dynamic allocation layer of the network to initially divide the priority factors of load units. The hierarchical load decision-making network adjusts the response strategy of each load unit in real time according to the load demand change and electricity price fluctuation; S62. In the dynamic allocation layer, analyze the current values of the real-time electricity price, load demand fluctuation, and response time to generate load priority factors; ； Among them, represents the priority factor of the load unit ; represents the weight for adjusting the priority during high demand fluctuations, represents the change rate of the electricity price response, represents the high - priority threshold of the electricity price, represents the demand fluctuation weight under medium demand fluctuation conditions, represents the electricity price smoothing coefficient, represents the median value of the electricity price, represents the response change rate for medium electricity price fluctuations, represents the weight for adjusting low - priority loads, represents the attenuation factor for the response time of controlling low - priority loads, represents the low - priority value of the electricity price, represents the response intensity of low - priority load electricity price fluctuations, represents the high - priority threshold of the response time, represents the maximum priority threshold of the response time, represents the high - fluctuation threshold of the load demand fluctuation, represents the low - fluctuation threshold of the load demand fluctuation, represents the load unit 's real - time electricity price, represents the load demand fluctuation of the load unit ; represents the load unit 's response time; S63. Classify the load units according to the priority factors, and divide the load into high-priority, medium-priority, and low-priority; S64. Generate a dynamic load management path in the hierarchical load decision-making network. Respond to high-priority load units first, process medium-priority load units when the electricity price is medium or the demand is stable, and arrange low-priority load units to be executed during low electricity price periods; S65. Monitor the execution of the load management path. If it is detected that the changes in the real-time electricity price, load demand fluctuation, or response time exceed the preset threshold during the path execution, recalculate the priority factors of each load unit and dynamically adjust the load management path.

Images